The Qualcomm® AI Hub Apps repository provides a collection of open-source sample applications and tutorials for deploying machine learning models on Qualcomm® devices. It targets developers looking to optimize on-device AI performance, offering recipes for various ML tasks and end-to-end workflow guidance.

How It Works

The apps leverage Qualcomm® AI Hub Models and support multiple runtimes including TensorFlow Lite, ONNX, and the Genie SDK for generative AI. Deployment targets are Android (API v30+) and Windows 11, with compute options spanning CPU, GPU, and NPU (Hexagon HTP). NPU acceleration requires specific Snapdragon chipsets and FP16 or INT8 precision.

Quick Start & Requirements

To get started, locate the desired OS and app within the repository's folders. Each app's README contains specific build and installation instructions. Supported deployment targets include Android 11 (API v30+) and Windows 11. NPU acceleration is optimized for Snapdragon chipsets (e.g., 8 Elite, 8 Gen 3/2/1, 888/888+).

Highlighted Details

• Supports TensorFlow Lite, ONNX, and Genie SDK runtimes.
• Optimized for on-device deployment on Android and Windows.
• NPU acceleration available on a wide range of Snapdragon chipsets.
• Includes sample apps for LLMs (GenAI), image classification, object detection, super-resolution, and speech-to-text.

Maintenance & Community

This repository is maintained by Qualcomm. Specific community channels or roadmaps are not detailed in the provided README.

Licensing & Compatibility

The Qualcomm® AI Hub Apps are licensed under BSD-3. This permissive license generally allows for commercial use and integration into closed-source projects.

Limitations & Caveats

While apps run on non-Snapdragon chipsets, NPU acceleration is exclusive to supported Qualcomm hardware. The README does not specify minimum hardware requirements beyond OS compatibility.